Skip to main content
GET
Python SDK

Authorizations

Authorization
string
header
required

Enter your API key (e.g. dai_sk_test_key_001)

Headers

Authorization
string

Query Parameters

category
string | null

Filter by category (e.g. 'tools', 'policies').

Example:

"tools"

project_id
string | null

Filter to a single project.

agent_name
string | null

Scope the menu to one agent (Use/Fork resolver) — agent-scope Use rows for that agent are included.

as_agent
boolean
default:false

Dashboard view of agent_name's list: resolve it with the identity the agent's own API key carries (taken from its latest run) instead of the caller's, record no routing decision, and return why each skill is on the list, the rows the caps cut, and the list the last run reported.

experiments
string | null

this_run when this answer shapes the run that asked: while the agent runs an experiment, the call gets an arm (a skill held out, or prompt B) and the response names the experiment. capable for a call that can route first but whose answer is not this run's: no arm, the experiment is named. Sent by decimalai 0.25+; leave it out to get the configured setup.

Pattern: ^(this_run|capable)$
evaluation
boolean
default:false

An evaluation run (decimalai.evaluate()): no experiment arm, the run is kept out of the agent's own results, and the response echoes evaluation. Implied by prompt_version and exclude_skills. Needs agent_name.

prompt_version
integer | null

An evaluation run on this saved version of the agent's prompt instead of the served one: its text comes back under prompt. 422 when the agent has no such version.

Required range: x >= 1
exclude_skills
string[] | null

An evaluation run without these skills (repeat the parameter per name): left out of every list, and refused to the run's load_skill calls. 422 for a name the agent does not have.

Cookies

decimal_session
string | null

Response

Successful Response